Highway tunnel construction safety monitoring method and system

Abstract

The invention discloses an expressway tunnel construction safety monitoring method and system, and belongs to the field of data processing.The expressway tunnel construction safety monitoring method comprises the following steps that a tunnel is logically divided into a plurality of continuous monitoring areas according to construction drawings, equipment positioning and process plans, and each independent area is divided into a plurality of continuous monitoring areas; the method has the beneficial effects that by constructing the multi-factor digital reference model library, lag judgment singly depending on a gas concentration threshold value is abandoned, and by monitoring a prospective index, namely the change rate of the gas concentration in real time, early warning can be immediately performed when the gas has an abnormal change trend due to the reduction of the ventilation efficiency; the absolute value of the gas concentration does not exceed the standard; the risk is identified before diffusion is completed in the early stage of danger accumulation, so that a time blind area from ventilation failure to traditional concentration threshold alarm is completely filled, and the timeliness of emergency response and the overall operation safety level are remarkably improved.

Description

Technical Field

[0001] This invention belongs to the field of data processing, and in particular relates to a method and system for monitoring the safety of highway tunnel construction. Background Technology

[0002] Long tunnel construction relies heavily on mechanical ventilation due to the enclosed space and lack of natural convection. The ventilation system must operate continuously to remove dust and harmful gases, maintaining operational safety. Current technology primarily relies on sensors installed within the tunnel to monitor gas concentrations such as methane, carbon monoxide, and oxygen, indirectly determining ventilation status. However, this protection mechanism suffers from significant latency: ventilation failures (such as fan malfunctions or duct damage) are not instantaneous, and the accumulation of harmful gases or a decrease in oxygen content requires time to diffuse to the sensor locations. A dangerous blind spot exists between ventilation cessation and the sensor reaching its alarm threshold. During this period, workers may not be able to detect abnormalities by sight and continue working, easily falling into dangerous situations such as sudden drops in visibility, oxygen deprivation, or even poisoning before the alarm is triggered. This leads to rushed emergency responses and significant safety hazards, necessitating improvement. Summary of the Invention

[0003] Therefore, it is necessary to provide a method and system for safety monitoring during highway tunnel construction to address the aforementioned problems.

[0004] The present invention is implemented as follows: a method for safety monitoring during highway tunnel construction includes the following steps: Based on the construction drawings, equipment positioning and process plan, the tunnel is logically divided into multiple continuously monitored areas (such as the working face area, material transportation channel, secondary lining area, etc.). For each independent area, based on the monitoring data under historical normal working conditions, combined with the environmental factors, working factors and time factors (such as geological conditions, construction activity type, shift handover period, etc.), a corresponding set of reference parameters θ_i is established for various working condition combinations and integrated into a digital reference model library. Based on environmental, operational, and time factors in each region, the corresponding reference parameter set θ_i for each region is called from the digital reference model library. Sensor data for each region is collected in real time. The real-time data stream of each region is synchronously compared and analyzed with the current reference parameter set θ_i of that region to determine whether the real-time data stream of each region deviates from the expected normal range or trend of the reference parameter set θ_i of the target region. When an anomaly is detected in a certain area, a graded early warning is automatically triggered based on the degree of deviation from the anomaly, and a targeted alarm is sent to the abnormal area and downstream affected areas (such as through audible and visual alarm devices, personnel positioning terminals, and dispatch centers) to clearly identify the risk location; at the same time, a preset adjustment command bound to the abnormal area is executed (such as adjusting the opening of the corresponding branch's air valve or starting the section standby fan).

[0005] In one embodiment, the present invention provides a method for safety monitoring during highway tunnel construction. The step of logically dividing the tunnel into multiple continuously monitored areas based on construction drawings, equipment positioning, and work plan, and for each independent area, establishing a corresponding set of reference parameters θ_i for various working condition combinations based on historical monitoring data under normal operating conditions, combined with environmental, operational, and time factors of that area, and integrating this set into a digital reference model library, specifically includes: Based on the construction drawings, equipment positioning, and work plan, the tunnel is logically divided into multiple continuously monitored zones, Zone_i (i=1, 2, ..., n). A dynamic feature vector F_i(t) is constructed for Zone_i to describe the state of the zone at time t. F_i(t) consists of three types of factors: environment, work, and time: F_i(t) = [E_i, W_i(t), T(t)], where: E_i is a static environmental factor (e.g., distance from the tunnel entrance, geological type, support type, which is relatively fixed during the construction phase); W_i(t) is a dynamic work factor (e.g., current activity type, number of equipment, personnel density), which changes in real time with the work process; and T(t) is a time factor (e.g., shift, date type, time since the last blast). The system calls upon monitoring data (such as time series of O2, CO, and dust concentrations) of Zone_i under historical normal ventilation conditions and performs correlation analysis with the feature vector F_i(t) of the same time period. Using machine learning algorithms (such as regression models or cluster analysis based on F_i(t), a corresponding gas concentration baseline parameter set θ_i is established for each feature state combination F_i. This parameter set includes: θ_i = {μ(F_i), σ(F_i), ΔC / Δt_max(F_i)}, where: μ(F_i) represents the expected baseline value (mean) of each gas concentration under feature F_i; σ(F_i) represents the allowable fluctuation range (standard deviation) of each gas concentration under feature F_i; ΔC / Δt_max(F_i) represents the maximum allowable rate of change threshold of each gas concentration under feature F_i. The baseline parameter set θ_i for all regions and all feature states is integrated into the digital baseline model library. When the target baseline parameter set θ_i is needed, the corresponding baseline parameter set θ_i is calculated and called from the digital baseline model library by matching or interpolation based on the current feature vector F_i(t) of each region obtained in real time.

[0006] In one embodiment, the present invention provides a method for monitoring the safety of highway tunnel construction, further comprising: For the connecting segment between two adjacent monitoring areas Zone_i and Zone_j (where j is an adjacent number of i), a composite feature vector is constructed for the connecting segment. At the same time, the baseline parameter sets θ_i and θ_j of the two adjacent areas are called for parallel comparison. The early warning trigger follows a dynamic fusion judgment mechanism, that is: if the real-time data exceeds the allowable fluctuation range or change rate threshold of either the baseline parameter set θ_i or θ_j, the environment of the connecting segment is determined to be abnormal.

[0007] In one embodiment, the present invention provides a method for monitoring the safety of highway tunnel construction, further comprising: Set up a whitelist region, and set a temporary model parameter tolerance window within the set time of the whitelist region. During the model parameter tolerance window, alarms (gas concentration exceeding the fluctuation range, whether the gas maximum change rate threshold is exceeded) of the specified gas in the whitelist region (such as gases directly caused by the process in this region, such as CO, dust, etc.) are blocked, or their alarm thresholds are temporarily adjusted to a relaxed value. After the set time ends, the original baseline parameter set θ_i under the current operating conditions of the whitelist region is restored.

[0008] In one embodiment, the present invention provides a method for monitoring the safety of highway tunnel construction, further comprising: The data, handling process, and environmental monitoring data within a specified time window after each early warning event are continuously recorded as case content. When the accumulated number of cases meets the batch update conditions, the case content is used to retrain and optimize the benchmark parameter set θ_i in the digital benchmark model library through machine learning algorithms.

[0009] In one embodiment, the present invention provides a highway tunnel construction safety monitoring system, comprising: The model library construction module is used to logically divide the tunnel into multiple continuously monitored areas (such as the working face area, material transportation channel, secondary lining area, etc.) based on construction drawings, equipment positioning and process plans. For each independent area, based on the monitoring data of the area under historical normal working conditions, combined with the environmental factors, working factors and time factors of the area (such as geological conditions, construction activity type, shift handover period, etc.), a corresponding set of reference parameters θ_i is established for various working condition combinations and integrated into a digital reference model library. The regional anomaly judgment module is used to call the corresponding benchmark parameter set θ_i of each region from the digital benchmark model library based on environmental factors, work factors and time factors of each region, collect sensor data of each region in real time, and compare and analyze the real-time data stream of each region with the current benchmark parameter set θ_i of that region to determine whether the real-time data stream of each region deviates from the expected normal range or trend of the benchmark parameter set θ_i of the target region. The early warning and command execution module is used to automatically trigger graded early warnings based on the degree of abnormal deviation when an anomaly is detected in a certain area. It sends targeted alarms to the abnormal area and downstream affected areas (such as through audible and visual alarm devices, personnel positioning terminals, and dispatch centers) to clearly identify the risk location. At the same time, it executes preset adjustment commands bound to the abnormal area (such as adjusting the opening of the corresponding branch's air valve or starting the section standby fan).

[0010] In one embodiment, the present invention provides a highway tunnel construction safety monitoring system, wherein the model library construction module includes: The region division and vector unit construction are used to logically divide the tunnel into multiple continuously monitored zones, Zone_i (i=1, 2, ..., n), based on construction drawings, equipment positioning, and work plan. A dynamic feature vector F_i(t) is constructed for Zone_i to describe the state of the zone at time t. F_i(t) consists of three types of factors: environment, work, and time: F_i(t) = [E_i, W_i(t), T(t)], where: E_i is a static environmental factor (e.g., distance from the tunnel entrance, geological type, support type, which is relatively fixed during the construction phase); W_i(t) is a dynamic work factor (e.g., current activity type, number of equipment, personnel density), which changes in real time with the work process; and T(t) is a time factor (e.g., shift, date type, time since the last blast). The benchmark parameter set establishment unit is used to call the monitoring data of Zone_i under historical normal ventilation conditions (such as the time series of O2, CO, and dust concentrations) and perform correlation analysis with the feature vector F_i(t) of the same time period. Through machine learning algorithms (such as regression models or cluster analysis based on F_i(t), a corresponding gas concentration benchmark parameter set θ_i is established for each feature state combination F_i. This parameter set includes: θ_i = {μ(F_i), σ(F_i), ΔC / Δt_max(F_i)}, where: μ(F_i) represents the expected benchmark value (mean) of each gas concentration under feature F_i; σ(F_i) represents the allowable fluctuation range (standard deviation) of each gas concentration under feature F_i; ΔC / Δt_max(F_i) represents the maximum allowable rate of change threshold of each gas concentration under feature F_i. The benchmark parameter set integration unit is used to integrate the benchmark parameter set θ_i under all regions and all feature state combinations into the digital benchmark model library. When the target benchmark parameter set θ_i is needed, the corresponding benchmark parameter set θ_i is called from the digital benchmark model library by matching or interpolating according to the current feature vector F_i(t) of each region obtained in real time.

[0011] In one embodiment, the present invention provides a highway tunnel construction safety monitoring system, which further includes: The connection segment fusion judgment module is used to construct a composite feature vector for the connection segment of two adjacent monitoring areas Zone_i and Zone_j (j is the adjacent number of i). At the same time, it calls the benchmark parameter sets θ_i and θ_j of the two adjacent areas for parallel comparison. The early warning trigger follows a dynamic fusion judgment mechanism, that is: if the real-time data exceeds the allowable fluctuation range or change rate threshold of either the benchmark parameter set θ_i or θ_j, the connection segment environment is judged to be abnormal.

[0012] In one embodiment, the present invention provides a highway tunnel construction safety monitoring system, which further includes: The whitelist alarm adjustment module is used to set whitelist areas. Within a set time period of the whitelist area, a temporary model parameter tolerance window is set. During the model parameter tolerance window period, alarms (gas concentration exceeding the fluctuation range, whether the gas maximum change rate threshold is exceeded) of specified gases in the whitelist area (such as gases directly caused by the process in this area, such as CO, dust, etc.) are blocked, or their alarm thresholds are temporarily adjusted to a relaxed value. After the set time ends, the original baseline parameter set θ_i under the current operating conditions of the whitelist area is restored.

[0013] In one embodiment, the present invention provides a highway tunnel construction safety monitoring system, which further includes: The benchmark parameter set optimization module is used to continuously record the data, handling process, and environmental monitoring data within a specified time window after each early warning event as case content. When the accumulated number of cases meets the batch update conditions, the benchmark parameter set θ_i in the digital benchmark model library is retrained and the parameters are optimized using the case content and machine learning algorithms.

[0014] Compared with existing technologies, the beneficial effects of this invention are as follows: By constructing a multi-factor digital benchmark model library, this invention abandons the lagging judgment that relies solely on gas concentration thresholds. By monitoring the rate of change of gas concentration in real time, a forward-looking indicator, it can immediately issue an early warning when abnormal gas change trends occur due to decreased ventilation efficiency, even if the absolute value of gas concentration has not yet exceeded the standard. This enables the identification of risks in the early stages of hazard accumulation and before diffusion is complete, thereby completely filling the time blind spot between ventilation failure and traditional concentration threshold alarms, significantly improving the timeliness of emergency response and the overall level of operational safety. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the first part of a method for monitoring the safety of highway tunnel construction, provided in an embodiment of the present invention.

[0016] Figure 2 This is a schematic diagram of the process for obtaining a digital benchmark model library provided in an embodiment of the present invention.

[0017] Figure 3 This is a schematic diagram of the second part of a method for monitoring the safety of highway tunnel construction, provided in an embodiment of the present invention.

[0018] Figure 4 This is a schematic diagram of the third part of a method for monitoring the safety of highway tunnel construction, provided in an embodiment of the present invention.

[0019] Figure 5 This is a schematic diagram of the fourth part of a method for monitoring the safety of highway tunnel construction, provided in an embodiment of the present invention.

[0020] Figure 6 This is a schematic diagram of the first part of a highway tunnel construction safety monitoring system provided in an embodiment of the present invention.

[0021] Figure 7 This is a schematic diagram of the model library construction module provided in an embodiment of the present invention.

[0022] Figure 8 This is a schematic diagram of the second part of a highway tunnel construction safety monitoring system provided in an embodiment of the present invention.

[0023] Figure 9 This is a schematic diagram of the third part of a highway tunnel construction safety monitoring system provided in an embodiment of the present invention.

[0024] Figure 10 This is a schematic diagram of the fourth part of a highway tunnel construction safety monitoring system provided in an embodiment of the present invention. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0026] In one embodiment, such as Figure 1 As shown, a method for safety monitoring during highway tunnel construction includes the following steps: Step S1: Based on the construction drawings, equipment positioning and process plan, the tunnel is logically divided into multiple continuously monitored areas (such as the working face area, material transportation channel, secondary lining area, etc.). For each independent area, based on the monitoring data under historical normal working conditions of that area, combined with the environmental factors, working factors and time factors of that area (such as geological conditions, construction activity type, shift handover period, etc.), a corresponding set of reference parameters θ_i is established for various working condition combinations and integrated into a digital reference model library. Step S2: Based on the environmental, operational, and time factors of each region, call the corresponding reference parameter set θ_i from the digital reference model library, collect sensor data of each region in real time, and compare and analyze the real-time data stream of each region with the current reference parameter set θ_i of that region to determine whether the real-time data stream of each region deviates from the expected normal range or trend of the reference parameter set θ_i of the target region. Step S5: When an anomaly is detected in a certain area, a graded early warning is automatically triggered according to the degree of deviation from the anomaly, and a directional alarm is sent to the abnormal area and the downstream affected area (such as through audible and visual alarm devices, personnel positioning terminals, and dispatch centers) to clearly identify the risk location; at the same time, the preset adjustment command bound to the abnormal area is executed (such as adjusting the opening of the corresponding branch air valve or starting the section standby fan).

[0027] Environmental factors (such as distance from the excavation face, geological gas inrush, and differences between completed and unsupported lining sections), operational factors (such as drilling and blasting operations at the excavation face, shotcreting operation areas, and areas with high vehicle traffic), and time factors (such as after blasting, during shift changes, and when equipment is started up in a concentrated manner) can all lead to drastically different gas generation rates, diffusion conditions, and background concentrations in different areas. A fixed global alarm threshold will either have a slow response in high-risk areas or frequent false alarms in safe areas. Step S1 establishes a dynamic digital benchmark model library, providing the system with a benchmark for intelligent judgment in the face of different environmental, operational, and time factors; Step S2 is responsible for real-time comparison to detect anomalies; and Step S5 is the automatic response execution. The advantage of this design is that it transforms the traditional passive and delayed gas concentration alarm into a proactive and forward-looking intelligent early warning and control system. It can significantly advance the identification point of ventilation failure from exceeding the concentration limit to when the gas concentration change begins to be abnormal, thereby systematically filling the blind spot of dangerous time and greatly shortening the emergency response time.

[0028] In one embodiment, such as Figure 2 As shown, a method for safety monitoring during highway tunnel construction includes step S1, which involves logically dividing the tunnel into multiple continuously monitored areas based on construction drawings, equipment positioning, and work plan. For each independent area, based on historical monitoring data under normal operating conditions, and considering environmental, operational, and time factors, a corresponding set of reference parameters θ_i is established for various combinations of operating conditions and integrated into a digital reference model library. This step specifically includes: Step S11: Based on the construction drawings, equipment positioning, and work plan, the tunnel is logically divided into multiple continuously monitored zones, Zone_i (i=1, 2, ..., n). A dynamic feature vector F_i(t) is constructed for Zone_i to describe the state of the zone at time t. F_i(t) consists of three types of factors: environment, work, and time: F_i(t) = [E_i, W_i(t), T(t)], where: E_i is a static environmental factor (e.g., distance from the tunnel entrance, geological type, support type, which is relatively fixed during the construction phase); W_i(t) is a dynamic work factor (e.g., current activity type, number of equipment, personnel density), which changes in real time with the work process; T(t) is a time factor (e.g., shift, date type, time since the last blast). Step S12: Call the monitoring data of Zone_i under historical normal ventilation conditions (such as the time series of O2, CO, and dust concentrations) and perform correlation analysis with the feature vector F_i(t) of the same time period (i.e., the time period corresponding to the acquisition of these monitoring data). Use machine learning algorithms (such as regression models or cluster analysis based on F_i(t)) to establish a corresponding gas concentration benchmark parameter set θ_i for each feature state combination F_i. This parameter set includes: θ_i = {μ(F_i), σ(F_i), ΔC / Δt_max(F_i)}, where: μ(F_i) represents the expected benchmark value (mean) of each gas concentration under feature F_i; σ(F_i) represents the allowable fluctuation range (standard deviation) of each gas concentration under feature F_i; ΔC / Δt_max(F_i) represents the maximum allowable rate of change threshold of each gas concentration under feature F_i. Step S13: Integrate the benchmark parameter set θ_i under all regions and all feature state combinations into the digital benchmark model library. When the target benchmark parameter set θ_i is needed, calculate and call the corresponding benchmark parameter set θ_i from the digital benchmark model library by matching or interpolation based on the current feature vector F_i(t) of each region obtained in real time.

[0029] The core design of steps S11-S13 is to structure and digitize the complex tunnel environment: Step S11 uses feature vector F_i(t) to uniformly quantify the multidimensional factors (environment, operation, time) affecting gas distribution; Step S12 uses historical data to learn and establish normal benchmark parameters under different operating conditions; Step S13 forms a digital benchmark model library that can be called in real time. The advantage of this design is that judging whether the gas is qualified is no longer a fixed value, but can dynamically reflect the normal standard in different areas and under different operating conditions, providing a highly customized and adaptive judgment basis for accurate anomaly identification, fundamentally solving the false alarm and missed alarm problems caused by fixed thresholds.

[0030] In one embodiment, such as Figure 3 As shown, a method for safety monitoring during highway tunnel construction also includes: Step S3: For the connecting segment of two adjacent monitoring areas Zone_i and Zone_j (j is a number adjacent to i), a composite feature vector is constructed for the connecting segment. At the same time, the reference parameter sets θ_i and θ_j of the two adjacent areas are called for parallel comparison. The early warning trigger follows a dynamic fusion judgment mechanism, that is: if the real-time data exceeds the allowable fluctuation range or change rate threshold of either the reference parameter set θ_i or θ_j, the environment of the connecting segment is determined to be abnormal.

[0031] Furthermore, based on gas diffusion models, the potential impact of upstream anomalies on downstream connecting sections can be assessed, enabling early warning. This mechanism ensures the continuity and security of monitoring at physical or logical boundaries, avoiding potential oversights due to regional divisions.

[0032] Step S3 is specifically designed to address potential monitoring blind spots arising from area division. By constructing composite feature vectors for connecting sections and employing a dynamic fusion mechanism of parallel comparison and alarm at higher elevations, it ensures seamless and conservative connection of judgment criteria at physical or logical boundaries. Its advantages include eliminating no-man's-land between areas, preventing early warning omissions due to switching of the reference parameter set θ_i or gas diffusion lag, and achieving continuous and comprehensive longitudinal safety monitoring of the tunnel.

[0033] In one embodiment, such as Figure 4 As shown, a method for safety monitoring during highway tunnel construction also includes: Step S4: Set a whitelist area. Set a temporary model parameter tolerance window within the set time of the whitelist area. During the model parameter tolerance window, the alarms (gas concentration exceeding the fluctuation range, whether the gas exceeds the maximum rate of change threshold) of the specified gas in the whitelist area (such as gases directly caused by the process in this area, such as CO, dust, etc.) are blocked, or their alarm thresholds are temporarily adjusted to a relaxed value. After the set time ends, the original reference parameter set θ_i under the current operating conditions of the whitelist area is restored.

[0034] This step aims to address a specific operational dilemma: compliant but high-emission short-term processes (such as welding, temporary start-up of internal combustion equipment, and other known compliant operations that generate large amounts of localized exhaust gases in a short period) can trigger invalid alarms, which, over time, will reduce personnel trust in the system. This whitelist mechanism is designed to reduce invalid alarms caused by predictable, non-ventilation system malfunctions, thereby improving system reliability.

[0035] In one embodiment, such as Figure 5 As shown, a method for safety monitoring during highway tunnel construction also includes: Step S6: Continuously record the data, handling process, and environmental monitoring data within a specified time window after each early warning event as case content. When the accumulated number of cases meets the batch update condition, use the case content to retrain and optimize the benchmark parameter set θ_i in the digital benchmark model library through machine learning algorithms.

[0036] This step gives the system the ability to continuously evolve. By using each early warning and response case as training data, the parameters are retrained periodically. The benefit of this design is that the system can self-calibrate as engineering progresses (such as geological changes and process adjustments), continuously converging and optimizing the normal benchmarks for each region, making the early warnings increasingly accurate.

[0037] In one embodiment, such as Figure 6 As shown, a highway tunnel construction safety monitoring system includes: Model library construction module 1 is used to logically divide the tunnel into multiple continuously monitored areas (such as the working face area, material transportation channel, secondary lining area, etc.) based on construction drawings, equipment positioning and process plans. For each independent area, based on the monitoring data of the area under historical normal working conditions, combined with the environmental factors, working factors and time factors of the area (such as geological conditions, construction activity type, shift handover period, etc.), a corresponding set of reference parameters θ_i is established for various working condition combinations and integrated into a digital reference model library. The regional anomaly judgment module 2 is used to call the corresponding benchmark parameter set θ_i of each region from the digital benchmark model library based on the environmental factors, working factors and time factors of each region, collect sensor data of each region in real time, and compare and analyze the real-time data stream of each region with the current benchmark parameter set θ_i of that region to determine whether the real-time data stream of each region deviates from the expected normal range or trend of the benchmark parameter set θ_i of the target region. The early warning and command execution module 5 is used to automatically trigger a graded early warning based on the degree of abnormal deviation when an anomaly is detected in a certain area, and send targeted alarms to the abnormal area and downstream affected areas (such as through audible and visual alarm devices, personnel positioning terminals, and dispatch centers) to clearly identify the risk location; at the same time, it executes the preset adjustment commands bound to the abnormal area (such as adjusting the opening of the corresponding branch air valve or starting the section standby fan).

[0038] The core basis for determining the downstream affected area is the airflow direction and gas diffusion model of the tunnel ventilation. First, the main airflow direction is determined based on the ventilation network design, and then the actual airflow path is verified using real-time wind speed sensor data. When an area is identified as an abnormal pollution source, the migration range and arrival time of the pollution cloud under the influence of airflow are estimated in real time using a simplified diffusion model based on the current wind speed, gas properties, and abnormal concentration. Simultaneously, sensor data from adjacent downstream areas is continuously monitored for abnormal trends as verification. Only areas located downwind of the abnormal source and within the estimated affected area, or those already showing data anomalies, are identified as downstream affected areas and receive targeted alarms, thus ensuring the accuracy and effectiveness of the early warning system.

[0039] In one embodiment, such as Figure 7As shown, a highway tunnel construction safety monitoring system includes a model library construction module 1 comprising: The region division and vector construction unit 11 is used to logically divide the tunnel into multiple continuously monitored zones, Zone_i (i=1, 2, ..., n), based on construction drawings, equipment positioning, and work plan. A dynamic feature vector F_i(t) is constructed for Zone_i to describe the state of the zone at time t. F_i(t) consists of three types of factors: environment, work, and time: F_i(t) = [E_i, W_i(t), T(t)], where: E_i is a static environmental factor (e.g., distance from the tunnel entrance, geological type, support type, which is relatively fixed during the construction phase); W_i(t) is a dynamic work factor (e.g., current activity type, number of equipment, personnel density), which changes in real time with the work process; and T(t) is a time factor (e.g., shift, date type, time since the last blast). The benchmark parameter set establishment unit 12 is used to call the monitoring data of Zone_i under historical normal ventilation conditions (such as the time series of O2, CO, and dust concentrations) and perform correlation analysis with the feature vector F_i(t) of the same time period. Through machine learning algorithms (such as regression models or cluster analysis based on F_i(t), a corresponding gas concentration benchmark parameter set θ_i is established for each feature state combination F_i. The parameter set includes: θ_i = {μ(F_i), σ(F_i), ΔC / Δt_max(F_i)}, where: μ(F_i) represents the expected benchmark value (mean) of each gas concentration under feature F_i; σ(F_i) represents the allowable fluctuation range (standard deviation) of each gas concentration under feature F_i; ΔC / Δt_max(F_i) represents the maximum allowable rate of change threshold of each gas concentration under feature F_i. The benchmark parameter set integration unit 13 is used to integrate the benchmark parameter set θ_i under all regions and all feature state combinations into the digital benchmark model library. When the target benchmark parameter set θ_i is needed, the corresponding benchmark parameter set θ_i is called from the digital benchmark model library by matching or interpolating according to the current feature vector F_i(t) of each region obtained in real time.

[0040] The feature vector F_i(t) is a complete state vector describing the region Zone_i at a certain time t; it is a dynamic input value that changes over time. The feature state combination F_i is a static representation of a typical working condition with similar characteristics, summarized from all historical F_i(t), such as the fixed combination "drilling and blasting operation at the working face - early shift". The benchmark parameter set θ_i is the output result corresponding one-to-one with each feature state combination F_i. It is the statistical safety boundary (mean, fluctuation range, and rate of change threshold) of the gas concentration under this working condition, learned through machine learning analysis of a large amount of historical normal data under this F_i state. Therefore, the relationship among the three is: the dynamically acquired feature vector F_i(t) is used to match the closest static feature state combination F_i in the digital benchmark model library, and then the static benchmark parameter set θ_i bound to it is called as a precise benchmark for judging whether the data is abnormal at the current moment.

[0041] In one embodiment, such as Figure 8 As shown, a highway tunnel construction safety monitoring system also includes: The connection segment fusion judgment module 3 is used to construct a composite feature vector for the connection segment of two adjacent monitoring areas Zone_i and Zone_j (j is the adjacent number of i). At the same time, it calls the benchmark parameter sets θ_i and θ_j of the two adjacent areas for parallel comparison. The early warning trigger follows the dynamic fusion judgment mechanism, that is: if the real-time data exceeds the allowable fluctuation range or change rate threshold of either the benchmark parameter set θ_i or θ_j, the connection segment environment is judged to be abnormal.

[0042] For example, suppose the blasting operation zone Zone_3 is connected to its adjacent material transport channel Zone_4. Due to the blasting operation, Zone_3 has a more lenient threshold for the rate of change of carbon monoxide (ΔC / Δt_max(F_3)) (e.g., allowing for short-term rapid increases); while Zone_4, as a transport channel, has a more stringent threshold for the rate of change of carbon monoxide (ΔC / Δt_max(F_4)). When a sensor located upstream of Zone_4 (i.e., near the boundary of Zone_3) detects a rapid increase in carbon monoxide concentration within a short period, and although the rate of increase does not exceed the lenient standard of Zone_3, it exceeds the strict threshold of Zone_4 itself, the system will immediately determine that an anomaly has occurred in the connection section based on the standard of Zone_4 by comparing the two benchmark models in parallel, and trigger an early warning.

[0043] In one embodiment, such as Figure 9 As shown, a highway tunnel construction safety monitoring system also includes: The whitelist alarm adjustment module 4 is used to set a whitelist area. Within the set time of the whitelist area, a temporary model parameter tolerance window is set. During the model parameter tolerance window, alarms (gas concentration exceeding the fluctuation range, whether the gas maximum change rate threshold is exceeded) of the specified gas in the whitelist area (such as gases directly caused by the process in this area, such as CO, dust, etc.) are blocked, or their alarm thresholds are temporarily adjusted to a relaxed value. After the set time ends, the original reference parameter set θ_i under the current operating conditions of the whitelist area is restored.

[0044] Building upon this, when the model parameter tolerance window is enabled, the system can dynamically calculate and delineate the expected impact area of ​​the process emissions, based not only on time but also on the gas diffusion model and real-time wind speed and direction. The tolerance rule only applies within the delineated, limited spatial impact area. Once abnormal gas diffusion beyond the boundary of the impact area is detected, or downstream sensors detect an unexpected increase in concentration, the system will immediately determine that emissions are out of control or ventilation has failed, globally cancel the tolerance window, and trigger an emergency alarm.

[0045] In one embodiment, such as Figure 10 As shown, a highway tunnel construction safety monitoring system also includes: The benchmark parameter set optimization module 6 is used to continuously record the data, handling process, and environmental monitoring data within a specified time window after each early warning event as case content. When the accumulated number of cases meets the batch update conditions, the benchmark parameter set θ_i in the digital benchmark model library is retrained and the parameters are optimized using the case content and a machine learning algorithm.

[0046] For example, suppose that in a warning event, Zone_7 (Region 7) of a transportation corridor repeatedly approaches but does not exceed the original threshold ΔC / Δt_max(F_7) during periods of heavy vehicle traffic. However, actual investigation shows that ventilation is already in a critically inefficient state. The baseline parameter set optimization module 6 uses the complete data stream of this event, the finally confirmed inefficient ventilation condition, and the environmental recovery curve after the intervention as a case study. The system analyzes this case study through machine learning and finds that the original model underestimates the gas accumulation rate under heavy vehicle traffic conditions. Therefore, it automatically optimizes and reduces the threshold of ΔC / Δt_max(F_7) for this region under this dynamic working factor W_7(t), and correspondingly tightens the fluctuation range σ(F_7). Subsequently, the system can identify the early decline in ventilation efficiency with higher sensitivity under similar conditions, realizing model self-calibration and iterative warning capabilities.

[0047] It should be understood that although the steps in the flowcharts of the various embodiments of the present invention are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the various embodiments may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least a portion of the sub-steps or stages of other steps.

[0048] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0049] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

[0050] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0051] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A method for safety monitoring during highway tunnel construction, characterized in that, The method for monitoring the safety of highway tunnel construction includes the following steps: Based on the construction drawings, equipment positioning and process plan, the tunnel is logically divided into multiple continuously monitored areas. For each independent area, based on the monitoring data under historical normal working conditions, combined with the environmental factors, working factors and time factors of the area, a corresponding set of reference parameters θ_i is established for various working condition combinations and integrated into a digital reference model library. Based on environmental, operational, and time factors in each region, the corresponding reference parameter set θ_i for each region is called from the digital reference model library. Sensor data for each region is collected in real time. The real-time data stream of each region is synchronously compared and analyzed with the current reference parameter set θ_i of that region to determine whether the real-time data stream of each region deviates from the expected normal range or trend of the reference parameter set θ_i of the target region. When an anomaly is detected in a certain area, a tiered early warning is automatically triggered based on the degree of deviation from the anomaly, sending targeted alarms to the anomaly area and downstream affected areas to clearly identify the risk location; at the same time, a preset adjustment command bound to the anomaly area is executed.

2. The method for monitoring the safety of highway tunnel construction according to claim 1, characterized in that, The step of logically dividing the tunnel into multiple continuously monitored areas based on construction drawings, equipment positioning, and work plan, and for each independent area, establishing a corresponding set of reference parameters θ_i for various working condition combinations based on historical monitoring data under normal operating conditions, combined with environmental, operational, and time factors, and integrating them into a digital reference model library, specifically includes: Based on the construction drawings, equipment positioning, and work plan, the tunnel is logically divided into multiple continuously monitored zones, Zone_i. A dynamic feature vector F_i(t) is constructed for Zone_i to describe the state of the zone at time t. F_i(t) consists of three types of factors: environment, work, and time: F_i(t) = [E_i, W_i(t), T(t)], where: E_i is the static environment factor; W_i(t) is the dynamic work factor, which changes in real time with the work process; and T(t) is the time factor. The system calls upon monitoring data of Zone_i under historical normal ventilation conditions and performs correlation analysis with the feature vector F_i(t) of the same time period. A machine learning algorithm is used to establish a corresponding gas concentration baseline parameter set θ_i for each feature state combination F_i. This parameter set includes: θ_i = {μ(F_i), σ(F_i), ΔC / Δt_max(F_i)}, where: μ(F_i) represents the expected baseline value of each gas concentration under feature F_i; σ(F_i) represents the allowable fluctuation range of each gas concentration under feature F_i; and ΔC / Δt_max(F_i) represents the maximum allowable rate of change threshold of each gas concentration under feature F_i. The baseline parameter set θ_i for all regions and all feature states is integrated into the digital baseline model library. When the target baseline parameter set θ_i is needed, the corresponding baseline parameter set θ_i is calculated and called from the digital baseline model library by matching or interpolation based on the current feature vector F_i(t) of each region obtained in real time.

3. The method for monitoring the safety of highway tunnel construction according to claim 2, characterized in that, Also includes: For the connecting segment between two adjacent monitoring zones, Zone_i and Zone_j, a composite feature vector is constructed for the connecting segment. At the same time, the baseline parameter sets θ_i and θ_j of the two adjacent zones are called for parallel comparison. The early warning trigger follows a dynamic fusion judgment mechanism, that is: if the real-time data exceeds the allowable fluctuation range or change rate threshold of either the baseline parameter set θ_i or θ_j, the environment of the connecting segment is determined to be abnormal.

4. The method for monitoring the safety of highway tunnel construction according to any one of claims 1 to 3, characterized in that, Also includes: Set a whitelist region, and set a temporary model parameter tolerance window within the whitelist region for a set time period. During the model parameter tolerance window period, the alarms of the specified gases in the whitelist region are blocked, or their alarm thresholds are temporarily adjusted to a relaxed value. After the set time ends, the original reference parameter set θ_i under the current operating conditions of the whitelist region is restored.

5. The method for safety monitoring during highway tunnel construction according to claim 1, characterized in that, Also includes: The data, handling process, and environmental monitoring data within a specified time window after each early warning event are continuously recorded as case content. When the accumulated number of cases meets the batch update conditions, the case content is used to retrain and optimize the benchmark parameter set θ_i in the digital benchmark model library through machine learning algorithms.

6. A safety monitoring system for highway tunnel construction, characterized in that, include: The model library construction module is used to logically divide the tunnel into multiple continuously monitored areas based on construction drawings, equipment positioning and process plans. For each independent area, based on the monitoring data of the area under historical normal working conditions, combined with the environmental factors, working factors and time factors of the area, a corresponding set of benchmark parameters θ_i is established for various working condition combinations and integrated into a digital benchmark model library. The regional anomaly judgment module is used to call the corresponding benchmark parameter set θ_i of each region from the digital benchmark model library based on environmental factors, work factors and time factors of each region, collect sensor data of each region in real time, and compare and analyze the real-time data stream of each region with the current benchmark parameter set θ_i of that region to determine whether the real-time data stream of each region deviates from the expected normal range or trend of the benchmark parameter set θ_i of the target region. The early warning and command execution module is used to automatically trigger a graded early warning based on the degree of abnormal deviation when an anomaly is detected in a certain area, and send targeted alarms to the abnormal area and downstream affected areas to clearly identify the risk location; at the same time, it executes the preset adjustment commands bound to the abnormal area.

7. The highway tunnel construction safety monitoring system according to claim 6, characterized in that, The model library building module includes: The region division and vector unit construction are used to logically divide the tunnel into multiple continuously monitored zones (Zone_i) based on construction drawings, equipment positioning, and process plans. A dynamic feature vector F_i(t) is constructed for Zone_i to describe the state of the zone at time t. F_i(t) consists of three types of factors: environment, work, and time: F_i(t) = [E_i, W_i(t), T(t)], where: E_i is the static environment factor; W_i(t) is the dynamic work factor, which changes in real time with the process; and T(t) is the time factor. The benchmark parameter set establishment unit is used to call the monitoring data of Zone_i under historical normal ventilation conditions and perform correlation analysis with the feature vector F_i(t) of the same time period. Through machine learning algorithms, a corresponding gas concentration benchmark parameter set θ_i is established for each feature state combination F_i. This parameter set includes: θ_i = {μ(F_i), σ(F_i), ΔC / Δt_max(F_i)}, where: μ(F_i) represents the expected benchmark value of each gas concentration under feature F_i; σ(F_i) represents the allowable fluctuation range of each gas concentration under feature F_i; ΔC / Δt_max(F_i) represents the maximum allowable rate of change threshold of each gas concentration under feature F_i. The benchmark parameter set integration unit is used to integrate the benchmark parameter set θ_i under all regions and all feature state combinations into the digital benchmark model library. When the target benchmark parameter set θ_i is needed, the corresponding benchmark parameter set θ_i is called from the digital benchmark model library by matching or interpolating according to the current feature vector F_i(t) of each region obtained in real time.

8. The highway tunnel construction safety monitoring system according to claim 7, characterized in that, Also includes: The connection segment fusion judgment module is used to construct a composite feature vector for the connection segment between two adjacent monitoring areas Zone_i and Zone_j. At the same time, it calls the benchmark parameter sets θ_i and θ_j of the two adjacent areas for parallel comparison. The early warning trigger follows a dynamic fusion judgment mechanism, that is: if the real-time data exceeds the allowable fluctuation range or change rate threshold of either the benchmark parameter set θ_i or θ_j, the connection segment environment is judged to be abnormal.

9. The highway tunnel construction safety monitoring system according to any one of claims 6 to 8, characterized in that, Also includes: The whitelist alarm adjustment module is used to set a whitelist area. Within a set time period of the whitelist area, a temporary model parameter tolerance window is set. During the model parameter tolerance window period, alarms for specified gases in the whitelist area are blocked, or their alarm thresholds are temporarily adjusted to a relaxed value. After the set time ends, the original reference parameter set θ_i under the current operating conditions of the whitelist area is restored.

10. The highway tunnel construction safety monitoring system according to claim 6, characterized in that, Also includes: The benchmark parameter set optimization module is used to continuously record the data, handling process, and environmental monitoring data within a specified time window after each early warning event as case content. When the accumulated number of cases meets the batch update conditions, the benchmark parameter set θ_i in the digital benchmark model library is retrained and the parameters are optimized using the case content and machine learning algorithms.

Images